Posted by Chris Lundie from dialin55.eagle.ca (206.186.242.200)on April 02, 1998 at 00:24:40:
In Reply to: Re: V1000 + BigRRedCroc = VERY cheap Voodoo. Woohoo! Test results inside... posted by Stephen Edie on April 01, 1998 at 02:00:53:
Actually I think the big bottleneck is not CPU speed, but the RRedline calls. Right now Big RRed uses simple, easy to use, but slow RRedline calls. In the future, we could switch to very fast and efficient calls. (In fact, you can re-write a lot of RRedline stuff yourself to make it quicker) Of course a fast CPU wouldn't hurt. :)
The 3Dfx GL could outperform the native Rendition OpenGL, but I think we would have to disable certain features to do that (in effect, cheat) Interestingly, the 3Dfx GL is just a wrapper on top of Glide. The Rendition GL is not a wrapper, it uses microcode to support OpenGL more or less "natively". That's why the never-released RRedline Quake 2 wasn't much better...
Too bad Rendition won't release specs for writing microcode. Then we could implement Glide calls right on the chip. (cool!)